Venture Finance under Flexible Information Acquisition

Transcription

1 Venture Finance under Flexible Information Acquisition Ming Yang Duke University Yao Zeng Harvard University October, 2012 (Preliminary and Incomplete) Abstract This paper investigates the finance for venture projects through the lens of information acquisition. It highlights the role of venture investors as information experts, who are more capable to acquire information about the market prospects of entrepreneurs venture technologies and then help screen them by their financing decision. It employs and further develops the flexible information acquisition approach to explore the specific interplay between contract design and information acquisition. It predicts specific contracts between entrepreneurs and venture investors, which correspond to the convertible preferred stock and the standard debt in terms of the allocation of cash flow rights. These predictions and related comparative statics are consistent with empirical evidences. Keywords: venture projects, flexible information acquisition, security design JEL: D82, D86, G24, G32, L26 We thank Snehal Banerjee, Emmanuel Farhi, Simon Gervais, Barney Hartman-Glaser, Benjamin Hebert, Steven Kaplan, Arvind Krishnamurthy, Josh Lerner, Stephen Morris, Jonathan Parker, Raghu Rajan, Adriano Rampini, David Robinson, Paola Sapienza, Hyun Song Shin, Andrei Shleifer, Alp Simsek, S. Viswanathan, and Michael Woodford for helpful comments. All errors are ours.

2 1 Introduction This paper investigates the finance for venture projects through the lens of information acquisition. We highlight the role of venture investors as information experts. Specifically, they are more capable to acquire information about the uncertain market prospects of entrepreneurs venture technologies by their industry experience, and thus help screen in good venture projects by their financing decision. The main contribution of this paper is to explain why and how different venture projects are financed by different contracts, and to reveal the role of information in this process. Specifically, we show that the standard debt and the convertible preferred stock are two most typical contracts to finance venture projects, which is consistent with the empirical evidences. Moreover, we figure out several key factors that determine the form of the optimal contract by performing comparative statics of the venture contracts. These factors include the cost-benefit efficiency and risk of venture projects, as well as the difficulty in evaluating the market prospect of venture projects, which is measured by venture investors information cost. This paper not only offers new insights on the finance for venture projects, but also reveals the interplay between security design and information acquisition in a more systematic way. The specific framework of this paper focuses on the interaction between information acquisition and contract design in financing venture projects. The acknowledgement of information acquisition in innovations and entrepreneurship is dated back to Frank Knight (1929) and Joseph Schumpeter (1942). Both authors highlight information acquisition and provision as important jobs conducted by innovators and entrepreneurs. Nevertheless, modern research on innovation and entrepreneurship have gradually shifted attention to the role of information acquisition by investors. As surveyed by Da Rin, Hellmann and Puri (2011), although entrepreneurs may have some information advantage on their own venture technology, modern venture investors, like venture capitalists, angels and banks, often have more expertise in acquiring information about the uncertain market prospects of venture technologies by their industry experience. Investors further help screen in good venture projects by such information expertise. Besides numerous anecdotal evidences (c.f. Kaplan and Lerner (2010) for a review), recent empirical literature (Kerr, Lerner and Schoar, 2011; Chemmanur, Krishnan and Nandy, 2012) have also increasingly acknowledged the screening effects by various venture investors. Therefore, it is necessary and natural to view contracts between entrepreneurs and venture investors as an incentive scheme to induce optimal information acquisition by venture investors. Although contract and security design is the focus of modern research in innovation and entrepreneurial finance, existing theoretical literature mainly pay attention to the aspects of monitoring and moral hazard (Schmidt, 2003; Casamatta, 2003; Hellmann, 2006), refinancing and staging of finance (Admati and Pfleiderer, 1994; Bergemann and Hege, 1998; Conelli and Yosha, 2003; Repullo and Suarez, 2004), as well as allocation of control rights (Berglof, 1994; Hellmann, 1998; Kirilenko, 2001; de Bettignies, 2008), leaving the 1

3 role of screening and information acquisition largely untouched. Also, existing models often fail to deliver consistent theoretical predictions with real-world contracts between entrepreneurs and venture investors. For instance, as documented in Da Rin, Hellmann and Puri (2011), the most commonly used double moral hazard models are able to approximately predict the contract of convertible debt between entrepreneurs and venture capitalists, but these models cannot deliver the correct specific allocation of cash flow rights between the two parties. This paper, to the best of our knowledge, is the first to explicitly investigate the channel of information acquisition in the finance of venture projects and deliver specific theoretical predictions that are consistent with corresponding empirical evidences regarding the contracts between entrepreneurs and venture investors (Kaplan and Stromberg, 2003). The contribution of this paper stems from a new concept, flexible information acquisition, which matches the characteristics of our research target. From the finance perspective, different venture projects are essentially characterized by the different cash flows they generate, which further require different contracts to reconcile the interests of entrepreneurs and investors. Designing contracts shapes investors incentives of information acquisition, and thus calls for a sufficiently flexible characterization of information acquisition to capture the potential variety of incentives generated by various contracts. Hence, the traditional approach of exogenous information asymmetry is inadequate. Moreover, classic models of information acquisition also fail to hit our target, since they implicitly impose some rigid information acquisition technologies with restrictions on signals (cf. Veldkamp (2011) for a review). Because of such rigidity, the strategic interplay between contract design and information acquisition cannot be fully analyzed. In contrast, our new approach of flexible information acquisition following Yang (2012a, 2012b) allows agents to choose not only how much information, namely, the quantitative nature of information to acquire, but also the qualitative nature of their information. Concretely, flexible information acquisition switches the gear of rational inattention (Sims, 2003) from macroeconomic decision problems to microeconomic strategic interactions as a new focus, in which agents can actively use their information capacities to choose any information structure (i.e., conditional distribution of signal realization given the fundamental). Such information structure characterizes both quantitative and qualitative nature of information acquired, which is shaped by the underlying contracts. For instance, a debt holder would allocate her attention to bad states as only the default risk matters, while an equity investor wants to pay more attention to good states as she enjoys the upside. This strategic interplay between contract design and information acquisition can be captured by neither exogenous information asymmetry nor rigid information acquisition models. This paper delivers new insights regarding the finance for venture projects and their information essence. Specifically, we can formulate the specific contracts required to finance various venture projects that differ in nature. In terms of the specific allocation of cash flow rights, these contracts corresponds to the convertible preferred stock and the standard debt. Specifically, 2

4 when the entrepreneur s optimal contract induces the investor to acquire information, the optimal contract is the convertible preferred stock. When the entrepreneur s optimal contract does not induce the investor to acquire information, the optimal contract turns out to be a standard debt. Importantly, the shape of these contracts not only reflects that the entrepreneur wants to design such contracts that the investor is willing to acquire more information to help distinguish between good states and bad states, but also characterizes the detailed information acquired by the investor in financing the venture projects. Moreover, different venture projects are endogenously financed by different contracts and potentially different venture investors. This can be clearly seen from the evolution of these contracts according to the change of natures of venture projects. Such evolution could be investigated from both the cross-sectional perspective to focus on industry-wide comparisons and the time-series perspective to look into firms life-cycle effect of innovations and venture projects. Concretely, as the market prospects of venture projects vary, investors will reallocate their attention accordingly, which results in different contracts eventually employed in financing corresponding venture projects. The optimal contracts also change as the characteristics of investors vary, mainly in their information costs. This dimension allows our analysis to cover venture capitalists, angels, banks and other financial institutions. This paper complements to Yang (2012a), which focuses on the role of flexible information acquisition in a pure liquidity provision problem and highlights the optimality of securitized debt. In that model, a seller has an asset in-place and proposes an asset-backed security to a potential buyer to raise liquidity. The buyer acquires information before her purchase decision, which leads to endogenous adverse selection. Similarly, Yang (2012a) assumes no information asymmetry before bargaining, but the buyer has an expertise in acquiring information about the fundamental following flexible information acquisition. The buyer collects the most payoff-relevant information determined by the shape of the security, and thus the seller designs the security in order to induce the buyer to acquire information least harmful to the seller s interest. It is shown that issuing securitized debt is uniquely optimal in raising liquidity. A key feature driving the unique optimality of debt is that the seller s asset is already in-place and the underlying cash flow does not depend on the trade, which means the aggregate risk is fixed. These results not only resolves the obscurity in Dang, Gorton and Holmstrom (2011) s optimality of quasi-debts, but also offers a benchmark for this paper, where the aggregate risk is variable in the finance of venture projects. When the aggregate risk is fixed as in Yang (2012a), the central implication of flexible information acquisition is endogenous information asymmetry and adverse selection. On the contrary, in this paper, venture projects can only be initiated if they are adequately financed. When the underlying cash flows of venture projects depend on the consequence of liquidity provision and thus the aggregate risk is variable, adverse selection is no longer the focus. Instead, entrepreneurs want to design such a contract that venture investors have the incentive to acquire information that is favorable to entrepreneurs interests. Intuitively, as venture investors 3

5 are more capable than entrepreneurs to acquire information about the uncertain market prospects of venture technologies, entrepreneurs would like to take advantage of such information expertise to help them screen in good venture projects. Therefore, debt is not optimal when information acquisition is desirable, and the model implications on liquidity are also different. Moreover, this paper delivers clear implications on liquidity provision and the associated specific allocation of cash flow rights in various economies and financial markets. In this model, the aggregate risk is variable in the finance of venture projects, so that the circumstance could be viewed as a production economy or a primary financial market. In contrast, the circumstance of Yang (2012a), where the aggregate risk is fixed, could be viewed as an exchange economy or a secondary financial market. Such comparison further helps us to understand why different optimal contracts arise in different circumstances. Earlier mechanism design literature pertaining to information gathering also highlight such comparison between different economies or financial markets and suggest that the value of information would be different accordingly (Cremer and Khalil, 1992; Cremer, Khalil and Rochet, 1998a, 1998b). But these papers do not focus on the security design problem and cannot make specific predictions on the forms of contracts in different economies or financial markets. Together with Yang (2012a), this paper fills this gap by delivering specific forms of contracts for liquidity provision in different circumstances. This paper is also closely related to a series of theoretical work on information, screening and optimal security design. These models predict non-debt-like securities in various circumstances where information asymmetry may be favorable (Nachman and Noe, 1994; Fulghieri and Lukin, 2001; Inderst and Mueller, 2006; Hennessy, 2009; Chakraborty and Yilmaz, 2011; Dow, Goldstein and Guembel, 2011; Fulghieri, Garcia and Hackbarth, 2012). Compared to those papers, most of which focus on ex-ante information asymmetry or rigid information acquisition, our model delivers clearer interaction between contract design and the corresponding information asymmetry resulted from information acquisition. This enables us to capture the role of information more explicitly in characterizing the optimal contract between entrepreneurs and investors. Also, thanks to the technique of flexible information acquisition, our model is able to characterize the shapes of optimal securities more specifically. Through this line, our model also highlights the standard intuition of information sensitiveness that non-debt-like securities may encourage investors to acquire information and facilitate information aggregation. The rest of the paper is organized as follows. Section 2 specifies the environment of venture finance and elaborates the concept of flexible information. Our key results, the optimal contracts for financing venture projects, are solved and discussed in Section 3. Section 4 performs comparative statics on the optimal contracts. In the final section, we conclude and discuss possible directions for further research. If otherwise noted, all proofs are attached in the appendix. 4

6 2 Model We present our model of venture finance. This model does not aim to capture every aspect of the financing of venture projects. Instead, we make assumptions to highlight the analysis of security design and focus on the essence of flexible information acquisition, which will be rigorously characterized in this section. We will discuss these assumptions and their implications in detail. 2.1 Venture Finance Consider an economy with two dates, t = 0, 1, and a single consumption good (money). There are two agents in this economy: an entrepreneur and a deep-pocket investor. Both agents are risk neutral. Their utility function is the sum of consumptions over the two dates: u = c 0 + c 1, where c t denotes agent s consumption at date t. 1 We assume that the entrepreneur starts at date 0 with zero initial wealth, while the deep-pocket investor has large endowment at date 0. In what follows we use the subscripts E and I to indicate the entrepreneur and the investor, respectively. We consider the innovation technology owned by the entrepreneur and its finance. To initiate the venture project at date 0, the technology requires a capital investment k > 0. The project cannot be initiated partially. Hence, the entrepreneur has to borrow k from the investor, by selling a security to the investor at date 0. The security traded constructs the contract between the two agents, so that we take the terms security and contract to mean the same thing and use them exchangeably. We focus on the cash flow aspect of contracts and innovations. If initiated, the venture project generates a non-negative verifiable random cash flow θ at date 1. We specify the processes of security design and information acquisition. The payment of an asset-backed security at date 1 is a mapping s : R + R + such that s(θ) [0, θ] for any θ. At date 0, both agents have a common prior P on the future cash flow θ of the venture project, if initiated. The entrepreneur designs the security, and then proposes a take-it-or-leave-it offer to the investor at price k at date 0. Facing the offer, the investor acquires information about the future cash flow θ in the manner of rational inattention (Sims, 2003; Yang, 2012a, 2012b), and then decides whether to accept the offer. This characterizes the process of flexible information acquisition, where the information acquired is measured by reduction of entropy. The information cost per unit reduction of entropy is µ. The concept of flexible information acquisition will be further elaborated in the next subsection. The entrepreneur could initiate the venture project if and only if the investor accepts the offer. This highlights the variable aggregate risk as a key feature in the finance of innovations and venture projects. 1 To focus on the essence, we do not explicitly consider discounting. Our results and underlying mechanisms do not rely on the discount between the two periods. 5

7 The implicit assumptions in the setting above are justified to capture the key features in the finance of venture projects, especially from an information acquisition perspective. First, the entrepreneur owns the technology while the investor has money, which reflects the double coincidence of wants. In particular, to highlight the double coincidence of wants, we assume that the entrepreneur has no money at all. This implies that the entrepreneur always prefers the venture project to be initiated, which is consistent with real practices. Second, both agents have a common prior on the uncertain profitability of the venture project but only the investor is able to acquire further information, which highlights the investor s information advantage in understanding market prospects. To focus on the essence of information acquisition, the bargaining process is abstracted as a take-it-or-leave-it offer from its reality, and no exogenous information asymmetry is assumed. This also implies that the allocation of control rights is not our focus. It is also worth highlighting what aspects of the finance of venture projects are abstracted away from the model, and to what extent these aspects affect our work. First, venture investors are unlikely to be risk neutral in reality. Nevertheless, the assumption of risk neutrality enables us to first focus on the role of information acquisition rather than risk allocation, which is of less interest for our purpose. Actually, our results would be only strengthened if we take investors risk aversion into account, which will be illustrated later. So we follow the risk neutral assumption to streamline our presentation and focus on the essence. Second, an interim cash flow, as modeled by the random variable θ, is unlikely to play a key role in the real-world venture finance setting. For example, most of the literature suggests that venture capitalist-backed firms lose money until well after going public and the exiting of the venture capitalists (Kaplan and Lerner, 2010; Da Rin, Hellmann and Puri, 2011). We could also, however, interpret the cash flow as an ex-post one, which already takes the consequence of venture investors exiting into account. In other words, the exiting nature of venture investors is already incorporated in the distribution of the future cash flow. This allows us to model a variety of venture investors not limited to venture capitalists, as well as to focus on the aspect of information acquisition. 2.2 Flexible Information Acquisition We follow Yang (2012a) to characterize the way through which the investor acquires information. In Yang (2012a), the author elaborates a general environment where an agent acquires information and thus makes a binary choice. The environment constructs the workplace of flexible information acquisition. 2 This environment and its corresponding implications also apply to our framework in this paper. Consider an agent who chooses a binary action a {0, 1} and receives a payoff u (a, θ), where θ Θ = R + is the fundamental, distributed according to a continuous probability measure P 2 For more detailed justification of this environment, see Woodford (2008) and Yang (2012a, 2012b). 6

8 over Θ. Before making the decision, the agent has access to a set of binary-signal information structures. 3 In particular, she observes binary signals x {0, 1} parameterized by a measurable function m : Θ [0, 1], where m (θ) is the probability of observing signal 1 if the true state is θ. When observing signal 1, the agent s optimal action is 1. The conditional probability function m (θ) describes the agent s decision rule of information acquisition. By choosing different functional forms of m (θ), the agent can make her signal covary with fundamental in any arbitrary way. Intuitively, for instance, if the agent s payoff is sensitive to fluctuations of the state within some range A Θ, she would pay more attention to this range by specifying m (θ) to be highly sensitive to θ A. This example also illustrates the concept of flexible information acquisition. As opposed to classic information acquisition technologies that often involve restrictions on the signals to be acquired, flexible information acquisition allows agents to choose signals drawn from any conditional distribution of the fundamental. The functional form of m(θ), intuitively, its shape, determines the quantity and quality of information acquired. We then characterize the quantitative and qualitative nature of information. As in Yang (2011a), information conveyed by an information structure m ( ) is defined as the expected reduction of uncertainty through observing the signal generated by m ( ), where the uncertainty associated with a distribution is measured by Shannon (1948) s entropy. Specifically, to formulate the process of flexible information acquisition and the underlying transition from prior to posterior, we employ the concept of mutual information, which is defined as the difference between agents prior entropy and expected posterior entropy: I (m) = H(prior) H (posterior) = g (E Θ [m(θ)]) ( E Θ [g (m (θ))]) = E Θ [g (m (θ)]) g (E Θ [m(θ)]), where g (x) = x ln x + (1 x) ln (1 x), and the expectation operator E Θ ( ) is with respect to θ under the probability measure P. In what follows we omit the subscript Θ for simplicity. We specify the cost of information. Denote M = {m L (Θ, P ) : θ Θ, m (θ) [0, 1]} as the set of binary-signal information structures. Let c : M R + be the cost of information. As in Yang (2012a), we assume that the cost is proportional to the associated mutual information: c (m) = µ I (m), 3 In general, an agent can get access to any information structure. But it could be shown that the agent always prefers binary-signal information structures in binary decision problems. See Woodford (2009) and Yang (2012b) for more discussions. 7

9 where µ > 0 is the marginal cost of information acquisition. This concludes the general environment of binary decision problem with flexible information acquisition. Built upon the concept of flexible information acquisition, the agent s problem is to choose a functional form of m(θ) to maximize her expected payoff minus the information cost. characterize the optimal decision rule m(θ) in the following proposition. We denote u(θ) = u(1, θ) u(0, θ), which is the the payoff gain of taking action 1 over action 0. We also assume that Pr [ u (θ) 0] > 0 to exclude the trivial case where the agent is always indifferent between the two actions. The proof is in Yang (2012a). Proposition 1. (Yang, 2012a) Given u, Θ, P, and µ, let m(θ) M be an optimal decision rule and p = E [m(θ)] be the corresponding unconditional probability of taking action 1. Then, i) the optimal decision rule is unique; ii) there are three cases for the optimal decision rule: a) p = 1, i.e., P r(m (θ) = 1) = 1 if and only if E [ exp ( µ 1 u (θ) )] 1 ; b) p = 0, i.e., P r(m (θ) = 0) = 1 if and only if c) 0 < p < 1 if and only if E [ exp ( µ 1 u (θ) )] 1 ; E [ exp ( µ 1 u (θ) )] > 1 and E [ exp ( µ 1 u (θ) )] > 1 ; (2.1) in this case, the optimal decision rule m(θ) is determined by the equation for all θ Θ, where u (θ) = µ (g (m (θ)) g ( p) ) (2.2) ( ) x g (x) = ln 1 x Proposition 1 fully characterizes the agent s possible optimal decisions of information acquisition. Case a) and case b) correspond to the scenario where the prior is extreme in the sense that there exists an ex-ante optimal action 1 or 0. These two cases do not involve information acquisition, and thus they also correspond to the scenario where the cost of information acquisition is sufficiently high. In contrast, case c), the more interesting one, involves information acquisition. Especially, the optimal decision rule m(θ) is not constant in this case, and neither action 1 nor action 0 is ex-ante optimal. Intuitively, this case corresponds to the scenario where the prior is 8. We

10 not extreme, or the cost of information acquisition is sufficiently low. In case c) where information acquisition is involved in the optimal decision rule, the agent equates the marginal benefit of information to the marginal cost of information. By doing so, the agent chooses the shape of her optimal decision rule m(θ) according to the shape of payoff gain u(θ) and her prior P, which process is consistent with the essence of flexible information acquisition. 4 3 Security Design and the Optimal Contract We consider the security design problem between the entrepreneur and the investor. We denote the entrepreneur s optimal contract by s (θ). The strategic circumstance between the entrepreneur and the investor is a dynamic Bayesian game with sequential moves. Concretely, the entrepreneur first designs the contract, and then the investor acquires information according to the contract and decides whether to accept the contract. In Proposition 1, we have characterized the general optimal decision rule for the agent who acquires information. Hence, we apply the results in Proposition 1 to the investor s decision problem, given the entrepreneur s contract, and then solve for the entrepreneur s optimal contract by backward induction. To distinguish from the general decision problem discussed above, we denote the investor s optimal decision rule of information acquisition as m s (θ), given the contract s(θ). The investor s optimal decision rule of information acquisition given the entrepreneur s optimal contract s (θ) will be denoted by m s(θ). We formally define the equilibrium of this model as follows. Definition 1. The sequential equilibrium is defined as a collection of the entrepreneur s optimal contract s (θ) and the investor s optimal decision rule of information acquisition m s(θ) based on which: i). Given u, Θ, P, k and µ, s (θ) and m s(θ) maximize the expected payoffs of the entrepreneur and the investor, respectively. ii). Both agents use the Bayes rule to update their beliefs about the fundamental θ, and follow sequential rationality. According to Proposition 1, there are three cases pertaining to the investor s behavior, given the entrepreneur s optimal contract. First, the investor may optimally choose to not acquire any information and accept the entrepreneur s optimal contract directly. Second, the investor may optimally acquire some information, induced by the entrepreneur s optimal contract, and then accept the entrepreneur s optimal contract with positive probability (but less than one). Third, the investor may directly reject the entrepreneur s optimal contract without any information acquisition. This last case corresponds to the outside option of the entrepreneur, who can always propose nothing to the investor and end up skipping the project. Therefore, it is less interesting to 4 See Woodford (2008), Yang (2012a, 2012b) for more examples on this decision problem. 9

11 characterize the last case explicitly. All the three cases could be accommodated by the equilibrium definition, and they may correspond to three types of equilibrium. In what follows, we first characterize the entrepreneur s optimal contract, focusing on the first two types of equilibrium. Specifically, we pay attention to the shape of the optimal contracts. As we will see below, the entrepreneur s optimal contracts in these two cases are different. This difference also implies different relationships between the shapes of contracts and the roles of information. We will later solve the entrepreneur s problem fully and deliver the exact optimal contract given certain exogenous parameters. 3.1 Optimal Contract without Inducing Information Acquisition In this subsection, we consider the case in which the entrepreneur s optimal contract is directly accepted by the investor without any information acquisition. In other words, it is optimal for the entrepreneur to design such a contract that does not induce the investor to acquire information. Concretely, this means P r [m s (θ) = 1] = 1. We first consider the investor s problem of information acquisition, given the entrepreneur s contract. Then we follow by characterizing the entrepreneur s optimal contract. Given a contract s(θ), the investor s payoff gain by accepting the contract over rejecting it is u I (θ) = u I (1, θ) u I (0, θ) = s (θ) k. (3.1) According to Proposition 1 and conditions (2.1) and (3.1), any contract s(θ) that is accepted by the investor without information acquisition must satisfy E [ exp ( µ 1 (s (θ) k) )] 1. (3.2) In particular, if the left hand side of the inequality (3.2) is strictly less than one, the entrepreneur could always lower s(θ) to some extent to increase her expected payoff gain, without changing the investor s incentive. Hence, this condition (3.2) always holds as an equality in equilibrium. Therefore, the entrepreneur s problem is to choose an contract s(θ) to maximize her expected payoff u E (s(θ)) = E [θ s(θ)] subject to the investor s information acquisition constraint E [ exp ( µ 1 (s (θ) k) )] = 1, the entrepreneur s individual rationality constraint E [θ s(θ)] 0, 10

12 and the feasibility condition 0 s(θ) θ. It is interesting to observe that the entrepreneur s individual rationality constraint is automatically satisfied, given the feasibility constraints. This comes from the fact that the entrepreneur is assumed to have no money at all to start. This fact also implies that the entrepreneur always prefers the project to be initiated, which is consistent with the real-world practices. Note that, it is not correct to interpret this in a way that the entrepreneur would like to contract with any individual venture investor, whatever the contract is. This is because we do not explicitly model the competition between different venture investors, so that the venture investor in our model indeed represents a collection of all venture investors in the market. As we would see, the entrepreneur s optimal contract in this case follows a standard debt. We analytically characterize this optimal contract by the following proposition, along with its graphical illustration. Proposition 2. If the entrepreneur s optimal contract s (θ) induces the investor to accept the contract without acquiring information in equilibrium, then it takes the form of a standard debt: s (θ) = min (θ, D ) where the face value D is determined by D = k µ ln(λ 1 µ), in which λ is a positive constant determined in equilibrium s(θ) θ Figure 1: The Unique Optimal Contract without Information Acquisition It is intuitive to have the standard debt as the optimal contract when it does not induces information acquisition. In the case without inducing information acquisition, the optimal contract indeed makes the investor break even between acquiring and not acquiring information. Hence, thanks to the definition of flexible information acquisition, any mean-preserving spread of the 11

13 optimal contract, which gives the entrepreneur the same expected payoff, would induce the investor to acquire information. Since it is optimal for the entrepreneur to design a less information sensitive security to deter information acquisition in this case, the debt is the least information sensitive one to provide the desired expected payoff to the seller. More specifically, it is the flat part of the debt that mitigates the investor s incentive to acquire information to the extent at which the buyer just wants to give up acquiring information, while delivers the highest possible expected payoff to the entrepreneur. As a result, the debt is the entrepreneur s uniquely optimal choice in this case. The standard debt, as the optimal contract, accounts for to the real-world scenarios in which some venture projects are financed by short-term fixed-income financial instruments. We will discuss this point in more details below. 3.2 Optimal Contract Inducing Information Acquisition In this subsection, we characterize the entrepreneur s optimal contract if it induces the investor to acquire information and accept the contract with positive probability (but less than one). In other words, it is optimal for the entrepreneur to design such a contract that induces the investor to acquire information. Concretely, this means P r [m s (θ) = 1] (0, 1). Again, according to Proposition 1 and conditions (2.1) and (3.1), any contract s(θ) that induces the investor to acquire information must satisfy E [ exp ( µ 1 (s (θ) k) )] > 1 (3.3) and E [ exp ( µ 1 (s (θ) k) )] > 1, (3.4) As discussed above, when conditions (3.3) and (3.4) are satisfied, neither accepting nor rejecting the contract is ex-ante optimal for the investor. As a result, the investor finds it optimal to first acquire some information and then make the decision according to her posterior. In other words, in this case, the contract proposed by the entrepreneur will induce information acquisition by the investor. Given such a contract s(θ), Proposition 1 and condition (2.2) also prescribe that the investor s optimal decision rule of information acquisition m s (θ) is uniquely characterized by s (θ) k = µ (g (m s (θ)) g (p s ) ), (3.5) where p s = E [m s (θ)] is the investor s unconditional probability of accepting the contract. In what follows, we also denote this unconditional probability induced by the entrepreneur s optimal contract by p s. Con- 12

14 dition (3.5) concludes the investor s decision problem, given the contract s(θ) proposed by the entrepreneur. We derive the entrepreneur s optimal contract by backward induction. Taking into account of investor s response m s, the entrepreneur chooses an contract s (θ) to maximize her expected payoff u E (s(θ)) = E [m s (θ) (θ s (θ))] (3.6) subject to (3.3), (3.4), 5 (3.5), the entrepreneur s individual rationality constraint E [m s (θ) (θ s (θ))] 0, and the feasibility condition 0 s (θ) θ. In characterizing the entrepreneur s optimal contract s (θ) when it induces information acquisition, we proceed by two steps. First, to streamline the presentation and fix the idea, we solve for an unconstrained optimal contract in this case without the feasibility condition 0 s(θ) θ. We denote the solution by ŝ(θ). We also denote the corresponding information acquisition rule by m s (θ). 6 We try to highlight that the feasibility condition is a mechanical restriction on the potential securities. Importantly, it is not directly relevant to the entrepreneur s motives in designing the contracts, which are aiming to induce the investor to acquire information that is most favorable to the entrepreneur. Hence, the unconstrained optimal contract ŝ(θ) helps to reveal the relationship between contract design and information acquisition in a clearer manner. After that, we resume the feasibility condition and characterize the exact optimal contract s (θ). This two-step approach streamlines our presentation and makes the intuition clearer. For the unconstrained optimal contract ŝ(θ) in the case with information acquisition, we have the following lemma. Lemma 1. In an equilibrium with information acquisition, the unconstrained optimal contract ŝ(θ) and its corresponding information acquisition rule m s (θ) are determined by ŝ (θ) k = µ (g ( m s (θ)) g (p s) ), (3.7) where and p s = E [m s(θ)], (1 m s (θ)) (θ ŝ(θ) + w ) = µ, (3.8) 5 It is worth noting that, according to Proposition 1, both conditions (3.3) and (3.4) should not be binding for the optimal contract; otherwise the investor would not acquire information. See Yang (2012a) for more details. 6 Note that, m s(θ) is not exactly the solution to the entrepreneur s problem without the feasibility condition, but is a translation of that solution. This will be seen clearer in the statement of Lemma 1. 13

15 where [ w = E (θ s g (p ] ( [ (θ)) s) g (p ]) 1 g (m 1 E s) s(θ)) g (m, s(θ)) in which p s and w are two constants that do not depend on θ. Lemma 1 exhibits the relationship between the unconstrained optimal contract and the corresponding decision rule for information acquisition. Recall Proposition 1, condition (3.7) specifies how the investor responses to the unconstrained optimal contract ŝ(θ) by adjusting her decision rule of information acquisition m s (θ). On the other hand, condition (3.8) is derived from the entrepreneur s optimization problem. It describes the entrepreneur s optimal choices of cash flows across states, which is summarized by the unconstrained optimal contract, given the investor s decision rule of information acquisition. In equilibrium, ŝ(θ) and m s (θ) are jointly determined. This again highlights the close relationship between security design and information acquisition in our context. Although it is mathematically difficult to solve the system of differential equations (3.7) and (3.8), we are able to deliver some important analytical characteristics of the unconstrained optimal contract ŝ(θ) and the corresponding information acquisition rule m s (θ). In particular, we are confident enough to speak to the shape of the unconstrained optimal contract ŝ(θ) as well as the actual optimal contract s (θ) by these analytical properties. The following lemma gives the key results. Lemma 2. In an equilibrium with information acquisition, the unconstrained optimal contract ŝ(θ) and the corresponding information acquisition rule m s (θ) satisfy ŝ (θ) θ = 1 m s (θ) (0, 1) (3.9) and m s (θ) θ = µ 1 m s (θ) (1 m s (θ)) 2. (3.10) We have several interesting observations from Lemma 2. First, condition (3.9) implies that the unconstrained optimal contract ŝ (θ) is strictly increasing. This is because, by definition, we have 0 < m s (θ) < 1, and thus the right hand side of (3.9) is positive. Second, the unconstrained optimal contract ŝ (θ) is strictly concave. This is because conditions (3.9) and (3.10) imply 2 ŝ(θ) θ 2 = µ 1 m s (θ) (1 m s (θ)) 2, the right hand side of which is negative. Therefore, the unconstrained optimal contract ŝ(θ) is an increasing concave function of θ. As opposed to the classic literature in security design, which often restrict the attention to non-decreasing securities (Innes, 1990; Nachman and Noe, 1994; DeMarzo and Duffie, 1999; DeMarzo 2005), it is intuitive to have the increasing unconstrained optimal contract as a result of our model. In the venture finance context, the investor provides 14

16 two different types of services to the entrepreneur. The first is the investment required to initiate the project, and the second is the information about the project s market prospect. In particular, these two services work in a way that they affect the probability of acceptance, at which the project takes place. As a result, it is natural for the entrepreneur to compensate the investor more in an event of higher profit, in exchange of these two types of services that make the project possible. More specifically, Lemma 2, especially the increasing and concave nature of the unconstrained optimal contract, reveals the relationship between the shape of the contract and the information acquired in a clearer way. Recall that the ratio m s (θ)/ θ represents the attention placed at state θ by the investor. When the slope of the contract changes more drastically around a state, which implies the change of payoff is more significant, the investor wants to acquire more information around that state. Furthermore, the contract is designed by the entrepreneur to take advantage of the investor s information expertise. Hence, the monotonicity and concaveness of the unconstrained optimal contract actually reflect the entrepreneur s payoff sensitiveness across states. On the one hand, the unconstrained optimal contract is strictly increasing, which always encourages the investor to acquire some information around all states. The reason is straightforward. As the entrepreneur cannot initiate the project by herself, she would like to encourage the investor to acquire information at all states in order to increase the probability of acceptance, at which the project would be initiated, to the maximal extent. On the other hand, the unconstrained optimal contract is concave, which implies that the information in all states is not equally valuable to the entrepreneur. In particular, more precise information is less valuable to the entrepreneur in a higher state than in a lower state. The intuition here is also clear. For a lower state, a given amount of additional information would significantly improve the posterior of the investor, which encourages her to make the decision of acceptance that is in favor of the entrepreneur. For a higher state, however, the same amount of additional information would be less valuable, because the higher state is already enough attractive to the investor given her prior, and thus a more precise posterior would not add much. In other words, it would not affect the probability of acceptance too much if the investor has less precise information for higher states, if she is confident enough by acquiring sufficiently precise information for the lower states. As a consequence, it is desirable for the entrepreneur to reduce the share to the investor in higher states. By doing this, the entrepreneur not only grasps a considerable share of the project s future cash flow, but also gives the investor enough and efficient incentives to acquire information that is favorable to the entrepreneur. Having characterizing the unconstrained optimal contract ŝ(θ), we bring in the feasibility constraints 0 s(θ) θ and analyze the actual optimal contract s (θ). It is instructive to have the following lemma to illustrate the possible relative positions between the unconstrained optimal contract and the feasibility constraints. 15

17 Lemma 3. Three possible relative positions between the unconstrained optimal contract ŝ(θ) and the feasibility constraints 0 s(θ) θ may occur in equilibrium, in the θ s space: i). ŝ(θ) intersects with the 45 line s = θ at ( θ, θ), θ > 0, and does not intersect with the horizontal axis s = 0. ii). ŝ(θ) goes through the origin (0, 0), and does not intersect with either the 45 line s = θ or the horizontal axis s = 0 for any θ 0. iii). ŝ(θ) intersects with the horizontal axis s = 0 at ( θ, 0), θ > 0, and does not intersect with the 45 line s = θ. In the three different cases, it is easy to imagine that the actual optimal contract s (θ) will be constrained by the feasibility condition in different ways. For example, the optimal contract will be constrained by the 45 line s = θ in case i) and by the horizontal axis s = 0 in case iii). Our concern is whether the presence of the two feasibility constraints would significantly change the interplay between security design and information acquisition, and thus the resulting optimal contract. Fortunately, the answer is no. As expected, the feasibility constraints are only mechanical. By imposing the feasibility conditions, we have the following characterization for the optimal contract: Lemma 4. In an equilibrium with information acquisition, the corresponding optimal contract s (θ) satisfies θ if ŝ(θ) > θ s (θ) = ŝ(θ) if 0 ŝ(θ) θ 0 if ŝ(θ) < 0 where ŝ(θ) is the corresponding unconstrained optimal contract., (3.11) Lemma 4 is important because it tells us how to construct an optimal contract with information acquisition from its corresponding unconstrained optimal contract. Concretely, the optimal contract s (θ) will follow its corresponding unconstrained optimal contract ŝ(θ) when the latter is within the feasible region 0 s θ. When the unconstrained optimal contract is out of the feasible region, the resulting optimal contract will follow one of the feasibility constraints that is binding. The expression of Lemma 4 is fairly simple but the result is not trivial. Importantly, the presence of the feasibility constraints indeed changes the shapes of the resulting optimal contracts from its unconstrained counterparts, which implies that the investor s incentives of information acquisition is also changed. Nevertheless, Lemma 4 ensures us that such change does not affect the entrepreneur s choice of the cash flow allocations in the states where the feasibility constraints are not binding. Also, in the states where the feasibility constraints are binding, Lemma 4 tells us that it is still optimal for the entrepreneur to just hit the binding constraints to exploit the investor s information advantage to the largest extent. 16

18 Therefore, we can apply Lemma 4 to the three cases of the unconstrained optimal contract ŝ(θ) described in Lemma 3. This gives the three potential cases of the optimal contract s (θ), respectively. Lemma 5. In an equilibrium with information acquisition, the optimal contract s (θ) may take one of the following three forms: i). When the corresponding unconstrained optimal contract ŝ(θ) intersects with the 45 line s = θ at ( θ, θ), θ > 0, we have s θ if 0 θ < θ (θ) = ŝ (θ) if θ θ ii). When the corresponding unconstrained optimal contract ŝ(θ) goes through the origin (0, 0), we have s (θ) = ŝ (θ) for θ R +. iii). When the corresponding unconstrained optimal contract ŝ(θ) intersects with the horizontal axis s = 0 at ( θ, 0), θ > 0, we have s 0 if 0 θ < θ (θ) = ŝ (θ) if θ θ We illustrate the three cases of optimal contracts s (θ) in the following graph... Figure 2: Potential Shapes of the Optimal Contracts The three potential cases of the optimal contract s (θ) take different shapes. Specifically, in case i), the optimal contract follows a standard debt in bad states but increases as a concave function in good states. In case iii), the payment of optimal contract in bad states is zero, while is an increasing and concave function in good states. Case ii) lies in between as a cut-off case, in which the payment of optimal contract is an increasing and concave function. Having characterizing the potential cases of the optimal contract s (θ) by its differential properties, we proceed by checking whether these three potential cases are indeed the valid solution to the entrepreneur s problem in an equilibrium with information acquisition. Interestingly, not all the three cases can occur in equilibrium. The following proposition tells us that only the optimal contract in case i) can actually sustain an equilibrium with information acquisition. 17

19 Proposition 3. If the entrepreneur s optimal contract s (θ) induces the investor to acquire information in equilibrium, then it must follow case i) in Lemma 5, which corresponds to a (participating) convertible preferred stock with a face value θ s(θ) θ Figure 3: The Unique Optimal Contract with Information Acquisition The underlying intuition for us to rule out the last two potential cases of the optimal contract is consistent with the practice of venture finance in reality. Recall that the investor can always reject the offer and enjoy her outside option, which is normalized to zero in the context, if her expected payoff is negative. The rejection of the contract, however, is always sub-optimal to the entrepreneur, as long as the project has a positive expected future cash flow. As a result, the entrepreneur wants to make sure that the investor, who makes the initial investment and takes cost to acquire information, is sufficiently compensated so that she is willing to accept the contract. The last two cases described in Lemma 5 corresponds to the scenarios in which the investor is not enough compensated so that the contract is rejected. To link this intuition to the real-world practice, because venture technologies often involve considerable intangible parts and high uncertainty, venture investors often ask for a large amount of compensations in bad states for a downside protection, in terms of the cash flow rights. Actually, the last two cases in Lemma 5 look closest to the common stock, which is the least used contract between entrepreneurs and investors in real-world venture finance cases (Kaplan and Stromberg, 2003; Kaplan and Lerner, 2010; Lerner, Leamon and Hardymon, 2012). In particular, Proposition 3 offers a very specific prediction on the entrepreneur s optimal contracts which may be employed in financing venture projects. Specifically, the optimal contract follows a debt in bad states until its face value and still increases in good states, which corresponds to a convertible preferred stock. In particular, it is closet to the participating convertible preferred stock, which grants the holder a right to receive both the face value and their equity participation as if the stock were converted, in the event of a sale or liquidation. This is consistent with the empirical evidences of venture capital contracts, in which the convertible preferred stock 18